Remote whisper.cpp engine fails with `Whisper server error: 404 File Not Found (/audio/transcriptions)`

I pointed Storyteller's whisper.cpp (remote) engine at a plain whisper.cpp whisper-server instance on another machine and every transcription fails with Whisper server error: 404 File Not Found (/audio/transcriptions). It isn't clear from the docs whether that engine requires the project's own remote transcription helper or whether any whisper-server will do.

Resolution: Any whisper.cpp whisper-server works — the helper daemon is a convenience wrapper, not a requirement — but Storyteller posts to the OpenAI-compatible path /audio/transcriptions, which whisper-server does not serve by default. Start whisper-server with --inference-path /audio/transcriptions and the 404 goes away. A working containerized example: podman run --security-opt=label=disable --device nvidia.com/gpu=0 -p 8080:8080 --rm -it -v /path/to/whisper_models:/models ghcr.io/ggml-org/whisper.cpp:main-vulkan "whisper-server --host 0.0.0.0 -m /models/ggml-base.bin --inference-path /audio/transcriptions". Two follow-on notes: the timing breakdown's upload line covers the entire remote transcription round trip, not just the network transfer, so a high percentage there is not automatically a network problem — but if transfers really are slow only from inside the container, suspect rootless container networking (pasta) rather than Storyteller. Also, whisper-server's --processors N (equivalent to Storyteller's "turbo mode") and Storyteller's parallelization setting do different things and should not both be raised above 1.